Help needed in a contact model (Abaqus/Standard)

[Izmar]

Hi everyone,

I am testing the effect of implant design variables (length, diameter, etc...) on stress in bone. I have been working on this project for several months and at this point I need someone to verify that my model is done properly.

I uploaded the model (Abaqus 6.9) to rapidshare. I would greatly appreciate any input.

http://rapidshare.com/files/314065414/model.rar.html

Thanks
Izmar

[Izmar]

***edited***

[Izmar]

Any input is greatly appreciated!

[Jorgen]

The file download was slow...
Can you explain your model or provide a few summary figures?

-Jorgen

[Izmar]

The file download was slow...
Can you explain your model or provide a few summary figures?

-Jorgen

I uploaded the model (ver 6.9) on another website, this one should go faster...

http://www.megaupload.com/?d=Q6UZWV84

Anyway, here's a summary of what I'm trying to do:

The model comprises of a screw inside a cubic bone block.
The aim is to test the effect of several screw design variables (length, diameter, shape, etc...) on the stress in bone.

Here's a summary of what I did so far:

- I modeled the screw geometry in Abaqus, used "cut geometry" to create a corresponding hole in the bone block and assigned material properties.
- I added a reference point on top of the screw head and used distributing coupling to link it to the surface. A concentrated force is applied to the ref point in the U1 direction
- The interaction between the screw and bone is defined as "finite sliding- adjust to remove overclosure", Default hard normal contact and 0.2 penalty friction tangential contact.
- In the load module, I applied the force to the ref point as mentioned before. The only BCs I applied were zero displacements to the front and back sides of the bone block (the block is not supposed to move). I did not apply any BCs to the screw itself or to the ref point.
- The screw dimensions are 6mm(length), 1.2mm(diameter), (0.1mm circular threads). The bone block is 20mm in all dimensions.
I used modified quad tetrahedral elements to mesh both parts.

I have the following questions:
1- Does everything I did so far make sense?
2- Is this the proper way to set the BCs for my model? someone suggested I should use kinematic coupling instead of distributing. It was also suggested to apply BCs to the ref point to restrain movements in all directions except U1, is this necessary?
3- My biggest problem so far is in the results and I think it has to do with my mesh. I am not sure what the smallest element size in the bone mesh should be. The screw thread is 0.1mm and has a sharp edge all around so I used 0.3 global seeds in the area of interest (close to the screw) and used seeding by edge on the edges of the bone block far from the screw. This resulted in a coarse mesh with about 10,000 elements. The problem is if I used smaller elements the results change drastically (max stress increases from 20 to 100MPa). I tried to use adaptive remeshing but it went on for several iterations but convergence was not achieved. I finally decided to decrease the global seeds to 0.05 which resulted in a model with 100,000+ elements and the analysis is still running so I am not sure what the results will be like. Is this because of the sharp edges in my model? is there any way around it?

I've been looking for help for months! I really appreciate your help

Thanks

[Jorgen]

1) Your approach sounds fine.
2) I personally use kinematic coupling more frequently. I would also add additional BCs to the ref point
3) That sounds like a mesh dependence problem. There is no easy way around the mesh density problem. If you need to know the max stress / strain, then you need to have elements that are sufficiently small that they can capture the gradients in the stress and strain fields. You can either be extra "smart" about the meshing, or just through a lot of elements at the problem and use a fast computer.

-Jorgen

[Izmar]

1) Your approach sounds fine.
2) I personally use kinematic coupling more frequently. I would also add additional BCs to the ref point
3) That sounds like a mesh dependence problem. There is no easy way around the mesh density problem. If you need to know the max stress / strain, then you need to have elements that are sufficiently small that they can capture the gradients in the stress and strain fields. You can either be extra "smart" about the meshing, or just through a lot of elements at the problem and use a fast computer.

-Jorgen

If I use kinematic coupling what dof should I restrain? I am not really concerned about the stress in the screw itself as much as the bone.

I guess my last problem is with the mesh. My mesh has to be fine as you mentioned since I have small surfaces and sharp edges involved in the contact, because of which there are stress concentration points at which the stress will keep increasing as I refine the mesh more. I am not interested in the stress cocentration areas as much as the stress field around the screw in general. How can I pick the most adequate element size for this purpose?

Thanks a lot for your help

Thanks a lot

[Izmar]

Here's a screenshot of my mesh.I know it is very coarse, but when I use finer elements the stresses become unrealistically high. Any suggestions would be greatly appreciated!

http://img689.imageshack.us/img689/3769/meshb.jpg

[Tompa]

It looks like you can calculate on a quarter of the model due to symmetry? Or even skip the cube and utilize axisymmetry?

Regarding the optimization, which parameters are you going to variate? I have worked pretty much with parametric optimization and parametric studies together with ABAQUS. Feel free to contact me through a response here or with a message, if you would like to have help with the analysis.

[Izmar]

It looks like you can calculate on a quarter of the model due to symmetry? Or even skip the cube and utilize axisymmetry?

Regarding the optimization, which parameters are you going to variate? I have worked pretty much with parametric optimization and parametric studies together with ABAQUS. Feel free to contact me through a response here or with a message, if you would like to have help with the analysis.

Tompa, thanks for your reply. Please check your PM